Visualizing Rous Sarcoma Virus Genomic RNA Dimerization in the Nucleus, Cytoplasm, and at the Plasma Membrane

Abstract

Retroviruses are unique in that they package their RNA genomes as non-covalently linked dimers. Failure to dimerize their genomes results in decreased infectivity and reduced packaging of genomic RNA into virus particles. Two models of retrovirus genome dimerization have been characterized: in murine leukemia virus (MLV), genomic RNA dimerization occurs co-transcriptionally in the nucleus, resulting in the preferential formation of genome homodimers; whereas in human immunodeficiency virus (HIV-1), genomic RNA dimerization occurs in the cytoplasm and at the plasma membrane, with a random distribution of heterodimers and homodimers. Although in vitro studies have identified the genomic RNA sequences that facilitate dimerization in Rous sarcoma virus (RSV), in vivo characterization of the location and preferences of genome dimerization has not been performed. In this study, we utilized three single molecule RNA imaging approaches to visualize genome dimers of RSV in cultured quail fibroblasts. The formation of genomic RNA heterodimers within cells was dependent on the presence of the dimerization initiation site (DIS) sequence in the L3 stem. Subcellular localization analysis revealed that heterodimers were present the nucleus, cytoplasm, and at the plasma membrane, indicating that genome dimers can form in the nucleus. Furthermore, single virion analysis revealed that RSV preferentially packages genome homodimers into virus particles. Therefore, the mechanism of RSV genomic RNA dimer formation appears more similar to MLV than HIV-1.

Keywords: FISH; RNA; RSV; dimerization; microscopy; retrovirus.

Figure 1

Three approaches to visualize genome dimerization in Rous Sarcoma virus (RSV). (A) Schematic of pRC.V8, the wild-type RSV proviral construct that forms the backbone plasmid that was modified to generate pRC.V8.GagCFP, which contains CFP fused to NC, disrupting the gagpol reading frame, resulting in a non-functional PR and pol (designated as pr* and pol*). pRC.V8.Gag-CFP contains insertion of either RNA binding stem-loops (MS2SL or BglSL) or a non-coding mCherry (ATG-mCherry) sequence. Below are schematics of the fluorophore-fused MS2 and Bgl binding proteins, which contain an NLS; (B) In the first approach, proviral constructs containing a noncoding mCherry sequence (pRC.GagCFP.ATG-mCherry) or PP7-derived sequence encoding RNA stem-loops (pRC.GagCFP.PP7) were co-transfected, each serving as a target for specific smFISH probes labeled with Alexa 647 (against mCherry; red) or Alexa555 (against PP7, green). WGA-Alexa 594 (gray) staining was used to outline the plasma membrane (image shown is a single z-plane); (C) In the second approach, proviral constructs containing the noncoding mCherry sequence (pRC.GagCFP.ATG-mCherry) were detected by smFISH probes labeled with Alexa 647 (red) or MS2 aptamer encoding RNA stem-loops (pRC.GagCFP.MS2) bound by the MS2-YFP protein (green). WGA-Alexa 647 staining was used to outline the plasma membrane (image shown is a single z-plane); (D) In the third approach, proviral constructs containing either the MS2 cassette (pRC.GagCFP.MS2) or Bgl RNA stem-loops (pRC.GagCFP.Bgl) were bound by either MS2-mCherry (red) or Bgl-YFP (green), respectively. WGA-Alexa 594 staining shown in gray outlined the plasma membrane (image is a single z-plane). In each vRNA labeling combination, only unspliced vRNAs were labeled. Representative microscopy images are shown for each condition. Images were obtained as z-stacks through the width of the cell using a 63× objective at 3.5× optical zoom. Foci were identified using the “Spot Function” in Imaris. Colocalization between RNA foci was determined as described in the Materials and Methods. Image histograms were adjusted in the Imaris software for ease of visualization; (E) The proviral constructs used were budding competent, as shown by immunoblotting using RSV anti-CA antibody to detect full-length Gag-CFP in cell lysates (left) and supernatants (right).

Figure 2

Examples of genome heterodimers located in the nucleus and at the plasma membrane. Shown are representative images of genome heterodimers within the nucleus (A–C), as defined by DAPI staining (blue), and at the plasma membrane, as defined by WGA-Alexa 594 staining (D,E) or Alexa 647 staining (F) in gray, for each of the described imaging methods. See Materials and Methods for further detail on the selection of the appropriate WGA conjugate. For the images showing genome heterodimers within the nucleus, the crosshairs expand into x,z and y,z planes to show the genome heterodimer is located within the DAPI staining in three dimensions (as outlined in white dashed line). For the images showing genome heterodimers at the plasma membrane, the two images on the left are a max projection image and inset showing the genome heterodimer at the plasma membrane (outline in white dashed line), while the two images on the right show the WGA-Alexa 594 or WGA-Alexa 647 staining (in gray) with Imaris Spots representing genome heterodimers. White arrowheads mark representative genome heterodimers at the plasma membrane (D,E) and at the plasma membrane and in the nucleus (F). Image histograms were adjusted in the Imaris software for ease of visualization for the purpose of this manuscript. Images were obtained using a 63× objective at a 3.5× optical zoom.

Figure 3

Genome heterodimer formation dependent on presence of dimerization initiation signal (DIS) (A) Schematic of pRC.V8 and pRC.V8.∆DIS proviral constructs, which contains a 77-nucleotide deletion (bp 219-296) removing the L3 stem-loop DIS. Each construct contains either 24x-MS2 RNA stem-loops or a non-coding mCherry RNA sequence after env; (B) Cells on the left were co-transfected with pRC.V8.ATG-mCherry, pRC.V8.MS2, and MS2-YFP (green) and subjected to mCherry RNA smFISH (red). On the right, cells were transfected with pRC.∆DIS.ATG-mCherry, pRC.∆DIS.MS2, and MS2-YFP (green) and mCherry RNA smFISH (red) was performed. Cells were imaged through the width of the nucleus, as assessed by DAPI staining, using a 63× objective at 3.5× optical zoom. The top rows of each panel show maximum projection images and the bottom row images indicate the location of foci identified using the Imaris Spots function; (C) Colocalization of dual-colored foci was analyzed using the Imaris Spots colocalization function (colocalization threshold 0.5 µm). Statistical significance was assessed by using negative binomial regression model, and a total of 30 cells were analyzed in two biological replicates, (*** p < 0.0001). Histograms were adjusted in the Imaris software for optimal visualization.

Figure 4

Single virion analysis in RSV shows preferential formation of genome homodimers: (A) Confocal microscopy of virions harvested from cell supernatant. QT6 fibroblasts were transiently transfected with proviruses containing MS2 and Bgl RNA stem-loop-containing and their respective fluorophore-labeled coat proteins. Visualized RNA genome foci are considered to be genome dimers, as they were present in virions harvested from the supernatant of transiently transfected cells. The top row shows the confocal microscopy images and the bottom row shows Spots labeled using the Imaris software. Images of virions were obtained using a 63× objective at 6× optical zoom; (B) Detection of Gag-CFP from virions harvested from cell supernatant expressing loop-containing provirus and their coat proteins by immunoblotting. The Gag-CFP is uncleaved, as expected, since CFP is located between NC and protease. pRC.V8, the wild-type parental proviral construct, which has normal protease activity, was used as a positive control for virus budding; (C) Quantitation of the number of genome homodimers and genome heterodimers present within harvested virions. Only RNA genomes that colocalized with Gag-CFP were analyzed, to ensure that RNA was located within a virion. A red-only or green-only focus was considered to be a genome homodimer and a red-green colocalized focus was considered to be a genome heterodimer. The colocalization threshold was set to 0.25 µm to determine red-green colocalization. To determine colocalization of an RNA focus with a Gag-CFP in virus particles, a colocalization threshold of 0.5 µm was used. Image histograms were adjusted in the Imaris software to optimize visualization.

Figure 5

Contribution of spliced vRNA to intracellular genome heterodimers in RSV: (A) Schematic of the provirus constructs used to visualize only unspliced vRNAs (top) or spliced and unspliced vRNAs (bottom). Each proviral construct contains either a MS2-24x loop cassette or a non-coding mCherry RNA sequence used for RNA smFISH. The location of the 5′ splice donor (5′ SD) and 3′ splice acceptor (3′ SA) are indicated; (B) Representative confocal microscopy images of genome heterodimers in cells co-transfected with the proviral constructs containing ATG-mCherry (red) and MS2-24x loop cassette and MS2-YFP (green) are shown here. Cells were imaged by a z-stack through the width of the cell, as determined by WGA membrane staining, using a 63× objective at 3.5× optical zoom. A colocalization threshold of 0.5 µm was used to determine whether two Spots were colocalized; (C) Comparison of the average number of heterodimers per cell between only unspliced vRNAs and spliced and unspliced vRNAs (n = 19 cells and n = 22 cells, respectively). The average number of heterodimers per cell is 3.79 ± 0.49 in unspliced only and 2.79 ± 1.10 in spliced and unspliced vRNAs. Significant outliers were identified and removed using the Grubbs’ test (p > 0.05, alpha = 0.5). No significant difference in the average number of genome heterodimers per cell was detected by unpaired t-test (p = 0.2132). Image histograms were adjusted in the Imaris software for optimal visualization.